You may have noticed that this week's lecture is short, and that's because there's a heavy load of important reading that you need to do. So before listening to this lecture you should read the Nuremberg Code, The Declaration of Helsinki, and The Belmont Report, all of which are available on the course website. Much of what we cover in this lecture will refer to those ethical codes. In section A, we're going to cover the idea of equipoise and we're going to talk about US regulation related to IRB review and approval, and then we're going to talk about privacy requirements of HIPAA. In clinical trials, in order to be ethically justified in randomly allocating participants to the treatment groups, we need to have a genuine uncertainty regarding the comparative therapeutic merits of the interventions being studied. In theoretical equipoise would imply that the evidence of benefits of each of the therapies are well balanced. But clinically, we define equipoise as a lack of consensus within the expert clinical community on the preferred therapy. So it's possible that individuals within the community may have an opinion that one treatment is better than the others, but what we need for clinical equipoise is for the clinical community as a whole to be in equipoise, not all the clinicians within the trial. Before we begin a trial, we have to make some sort of estimate regarding the difference in treatment effects in the groups that we plan to study. And we'll hear more about this is in the sample size lecture. But we make estimates about the difference in treatment effects, and these estimates are surrounded by some measure of uncertainty. So, in the figure, we show different sorts of estimates, and the uncertainty surrounding them. So the vertical line would indicate that there is no difference between the control therapy and the new therapy. And estimates to the right of the vertical line would indicate that the new therapy is better. Estimates to the left of the vertical line would indicate that the control therapy is better. The Xs are the estimate of the effect, and the arrows around the estimates represent our uncertainty. So in the top horizontal line, you can see that we think that the new therapy is better. And our estimates of uncertainty do not cross the vertical line. So they do not cross the null. And that means that we cannot ethically randomize people to receive either the new therapy or the control therapy because enough evidence already exists to make us believe that the new therapy is better than the control therapy. In the second horizontal line, we have the same estimated effect, meaning that we think the therapy is as good as we thought it was in the first horizontal line. But now we have much wider un, level of uncertainty around that estimate. And in fact, we are not sure that the new therapy is better than the control therapy, and we still have some uncertainty about which therapy is better. So, in the second horizontal line, we are in the state of equipoise. And finally, in the third horizontal line, we think that the new therapy is slightly better than the control therapy, but again our estimate of effects crosses the null so we are not sure. We don't have enough evidence yet to say that the new therapy is better than the control therapy, and we are in a state of equipoise. What we find in practice is that not all therapies show as much promise as we hoped when we were designing the study. And in fact, we are often quite optimistic about how efficacious a new therapy will be. It's also difficult to find a group of investigators that are all individually in a state of equipoise. This is why we require that there be a general state of equipoise in the clinical community. The investigators who feel strongly one way or the other are the ones that are the biggest advocates for the trial. And as trials go on we need a way to reevaluate equipoise during the trial. And this is called data monitoring, and we have a lecture on data monitoring later in the course. But this helps us determine if we are still in a state of equipoise during the trial and whether it's ethical to continue the trial. Regulatory law is codified in the code of federal regulations and published in the federal register. And here is a website where you can go and look at the code of federal regulations. Title 45, part 46, on the protection of human subjects. And this section applies to any institution that uses federal funds to conduct research on humans. If the institution uses any federal funds, then the regulation applies to all research done at that Institution. And it says that the institution has the responsibility for protecting the rights and welfare of human subjects. And it requires that the research be reviewed and approved by an institutional review board, or IRB, before the federal funding can be obtained. So, if you win an award with the NIH, they will not transfer any money to your institution until you can demonstrate that your study's been reviewed and approved by your IRB. The FDA has similar regulations for studies under its purview and those are codified in title 21, part 50, and title 21, part 56. The federal regulations give IRBs quite a bit of authority. IRBs can approve, require modifications to, or disapprove of research activities. They have to approve all consent documents, or they can, they can approve a waiver of consent when it's not practical to get consent before participation, such as emergency situations. They are also required by law to conduct continuing review of research at least once per year. And, during these reviews, they can suspend or terminate the approval of the research if it is not being performed in accordance with what was approved by the IRB, or if they find that there are unexpected undue risks associated with the treatments. In order to approve a research application, the IRB has to determine that all of the criteria on this list are met. They have to determine that the risks to the subjects are minimized and that they're reasonable in relationship to the anticipated benefits. They have to assure that the selections of subjects is done in an equitable manner, and that consent is obtained and documented before participation. They have to determine that there are adequate provisions to protect the privacy and the confidentiality of the participants, and that there are safeguards for vulnerable populations. And finally the last bullet is not actually mentioned in any of the ethical codes but is part of the regulation. The IRB has to review the provisions that are made for data monitoring. So submissions to the IRB have to describe and justify the research question and the rationale. They have to detail the procedures of the research. Who's going to do the research? Where is it going to be done? And they have to describe the risks that might be incurred by taking part in the research. Investigators also have to submit their recruitment procedures and their materials, so the IRB can determine if the materials appropriately represent the protocol to the participants. The basic elements of consent are also codified in the Code of Federal Regulations, and it says that the consenting process and the forms that are used to document the consent should inform patients of all of these elements. They have to describe the purpose of the research, including the risks and the benefits. They also have to describe what are the alternative courses of treatment if you choose not participate in the trial. They have to describe how the records are being kept confidential. And if there is compensation for injury, what is it? They also have to give contact information for the investigator and for the overseeing IRB. And they have to make it clear that participation in the research is entirely voluntary, and that the participants can withdraw at any time. The original purpose of the HIPAA regulation was to institute national standards for electronic health care transactions. And the intent was to improve the efficiency of these transactions. But it was recognized that this was referring to people's private data, and it was felt that it was important to put into law some guidance as to how the privacy of the data should be protected. And entities are required to follow HIPAA if they engage in any electronic transfer of health information. So this applies to the transfer of information in healthcare provider settings, but it also has implications for research settings, since investigators are also involved in the use and transfer of people's private medical data. In clinical trials, we use data in several different forms, and we have to describe who is allowed to see different sorts of data in either the consent form or a HIPAA form that is also signed by the participant before they begin the study. And the use of raw data is actually fairly limited, and by raw data I mean data that includes personally identifying information, such as name or social security number, or date of birth, or other identification that could be directly linked to a person. And in clinical trials, only auditors usually see this sort of data, in addition to the study staff at the clinical center that collect the data. So people auditing from the IRB, the study sponsor, be it NIH or a pharmaceutical company, the FDA, or even someone auditing from the coordinating center, might see the raw data at the clinical center. But what's usually written on the data collection forms and what's recorded in the database does not have personally identifying data on it. Instead, it's identified by a study code. So each participant, instead of recording in the database their name, we record a code such as AB004. And that code is used to identify that person on all their data collection forms. And a much broader group of people may see the coded data. The coordinating center collects and manipulates the coded data. The IRBs, the FDA, and the sponsors may see the coded data. The data monitoring, the data monitoring committees see coded data, and also collaborators from different sites may see coded data. In addition federally sponsored research has requirements that data be shared and made publicly available after the conclusion of the trial. So data is also seen in other forms, such as the identified data sets and limited use data sets, and these both make a much wider umbrella of people who may see coded data. De-identified data sets are those which, as much as possible, all potentially identifying information has been removed. And this is extremely restrictive. It includes all dates, including date of birth, date of death, date of diagnosis, all dates of study visits and it's extremely difficult to do, and it makes the resulting data sets of limited use. There are also data sets called limited use data sets, that have the obvious personal identifies removed, and can only be used with a data use agreement. So investigators have to receive IRB approval to use the data set, and then they have to sign an agreement saying that they will not share the data with other people, and that they will not try to identify the people in the data set.
